"michael" <miblogic@yahoo.com> writes:
> with regards to unique rows, i think we should deviate from the sql
> standard.
That's not happening.
> or if this isn't possible, we should at least document this in
> postgresql's unique index/unique constraint's gotchas, make more obvious.
It is documented ... you just quoted the docs.
> i'm really surprised that two rows with equal values is allowed
It is not the case that "NULL = NULL", so why would you think that the
two rows are equal?
You probably need to rethink what you are using NULL for in this table.
It's really only good for cases where you mean to indicate that a value
is unknown.
regards, tom lane